Analogue discrete-time basic structures for adaptive IIR filters

An analogue discrete-time implementation of an adaptive infinite-impulse-response (IIR) filter is presented. The filter structure uses the IIR modelling capability of first- and second-order filter sections and the adaptation process is a variable step-size version of the least-mean-squared (LMS) algorithm. The gradient signals were chosen to allow the use of the whole dynamic range of the four-quadrant analogue multiplier. With respect to implementation problems, the effects of nonidealities on the building elements are investigated and reduced using appropriate circuit techniques. The functional behaviour of the proposed structure is demonstrated for a low-complexity third-order adaptive filter with a parallel architecture.